Protein Kinase B/Akt Mediates Effects of Insulin on Hepatic Insulin-like Growth Factor-binding Protein-1 Gene Expression through a Conserved Insulin Response Sequence*

Insulin regulates the expression of multiple hepatic genes through a conserved insulin response sequence (IRS) (CAAAAC/TAA) by an as yet undetermined mechanism. Protein kinase B/Akt (PKB/Akt), a member of the PKA/PKC serine/threonine kinase family, functions downstream from phosphatidylinositol 3′-kinase (PI3K) in mediating effects of insulin on glucose transport and glycogen synthesis. We asked whether PKB/Akt mediates sequence-specific effects of insulin on hepatic gene expression using the model of the insulin-like growth factor binding protein-1 (IGFBP-1) promoter. Insulin lowers IGFBP-1 mRNA levels, inhibits IGFBP-1 promoter activity, and activates PKB/Akt in HepG2 hepatoma cells through a PI3K-dependent, rapamycin-insensitive mechanism. Constitutively active PI3K and PKB/Akt are each sufficient to mediate effects of insulin on the IGFBP-1 promoter in a nonadditive fashion. Dominant negative K179 PKB/Akt disrupts the ability of insulin and PI3K to activate PKB/Akt and to inhibit promoter activity. The IGFBP-1 promoter contains two IRSs each of which is sufficient to mediate sequence-specific effects of insulin, PI3K, and PKB/Akt on promoter activity. Highly related IRSs from the phosphoenolpyruvate carboxykinase and apolipoprotein CIII genes also are effective in this setting. These results indicate that PKB/Akt functions downstream from PI3K in mediating sequence-specific effects of insulin on the expression of IGFBP-1 and perhaps multiple hepatic genes through a conserved IRS.

The binding of insulin to its cell surface receptor results in the activation of multiple signaling pathways (1). Phosphatidylinositol 3Ј-kinase (PI3K) 1 plays an important role in medi-ating cellular effects of insulin (2,3). Rapamycin-sensitive pathways mediate PI3K-dependent effects of insulin on p70S6 kinase activity and cellular proliferation (4) and hexokinase II expression (5). Protein kinase B/Akt (PKB/Akt), a PKA/PKC serine/threonine kinase family member (6), is activated through a PI3K-dependent mechanism that is not blocked by rapamycin (7) and mediates effects of growth factors on cell survival (8 -10). PKB/Akt is thought to mediate the phosphorylation and inactivation of glycogen synthase kinase-3 by insulin (11) and stimulates glucose transporter translocation in 3T3-L1 adipocytes (12). The role of PKB/Akt in mediating sequence-specific effects of insulin on hepatic gene expression has not been determined.
Insulin regulates gene expression in a variety of tissues (13). Hepatic expression of insulin-like growth factor binding protein-1 (IGFBP-1) is rapidly suppressed by insulin at the level of gene transcription (14 -16) through two insulin response sequences located ϳ100 bp 5Ј to the RNA cap site (17,18). Each IRS is sufficient to mediate effects of insulin on promoter activity (18,19) and closely resembles IRSs, which mediate effects of insulin on several other genes in the liver, including phosphoenolpyruvate carboxykinase (PEPCK) (20) and apolipoprotein CIII (apoCIII) (21). These observations have suggested that a common, but as yet undetermined mechanism may mediate the effects of insulin on multiple hepatic genes.
Recent studies with wortmannin, a PI3K inhibitor, indicate that PI3K is necessary for insulin to reduce IGFBP-1 and PEPCK mRNA levels in liver-derived cells stimulated with dexamethasone and either glucagon or 8-(4-chlorophenylthio)-cAMP, respectively (22,23). Since multiple mechanisms involving distinct cis-acting DNA response sequences may mediate effects of insulin on gene expression in cAMP/glucocorticoidstimulated cells (24,25), we chose to examine mechanisms mediating the effects of insulin on IGFBP-1 expression under basal conditions where specific IRSs are known to be involved (18). To our knowledge, this study provides the first report that PKB/Akt is necessary and sufficient to regulate gene expression through a specific response sequence, and, in this case, mediates effects of insulin on basal hepatic gene expression through a conserved IRS.

MATERIALS AND METHODS
cDNA Clones and Plasmid Expression Vectors-The SauI-HgaI fragment of the IGFBP-1 promoter was inserted into the pGL2 polylinker (Promega) (BP1.luc), and an NheI site was created by site-directed mutagenesis of bp Ϫ112 and Ϫ113 (BP1.Nhe), as reported previously (18). The BamHI site in pGL2 was disrupted and the IGFBP-1 promoter containing the NheI site was subcloned into this vector. Using a unique BamHI site located 3Ј of IRSB in the IGFBP-1 promoter (Table I), we excised the NheI-BamHI fragment and replaced it with synthetic double-stranded oligonucleotides to create altered IRSs. All constructs were confirmed by dideoxy sequencing.
Cell Culture and Northern Blotting-HepG2 cells were grown in 100-mm dishes to ϳ70% confluence in Opti-MEM plus 5% BCS, rinsed, and stabilized for 24 h in 10 ml of DMEM with 1 g/liter BSA. LY294002 (Calbiochem) and wortmannin and rapamycin (Sigma) were added in dimethyl-d 6 -sulfoxide (Me 2 SO) (0.125%, v/v) 30 min prior to addition of recombinant human insulin (Sigma). Wortmannin or Me 2 SO was added again 90 min later to maintain levels (22). Total cellular RNA was prepared with TRI-Reagent-LS per the manufacturer's instructions (Molecular Research) and 10 g loaded for gel electrophoresis and transfer to Nytran membranes (Qiagen). Membranes were probed with IGFBP-1 and GAPDH cDNAs labeled with [␣-P 32 ]dATP (Amersham Pharmacia Biotech) by random priming and transcripts visualized by autoradiography and quantified by phosphorimaging.
Transient Transfection Assays-HepG2 cells (1 ϫ 10 6 /60-mm dish) were plated in Opti-Mem/BCS and transfected with calcium phosphate precipitates including a cytomegalovirus-driven ␤-galactosidase (␤-gal) expression vector, as before (35). Cells were rinsed, then refed with DMEM/BSA and appropriate supplements. Lysates were prepared 18 h later with lysis buffer (Promega). Luciferase activity was measured with an Optocomp-I luminometer (MGM Instruments) and results adjusted for ␤-gal activity. Each experiment was performed in triplicate and repeated at least three times.
Immunoprecipitation, Western Blotting, and Kinase Assays-Cell lysates were prepared in the presence of phosphatase and protease inhibitors (36) and cleared by centrifugation. Endogenous p70S6 kinase was precipitated with anti-p70 antiserum (Upstate Biotechnology, Inc.) and protein-G and kinase activity was measured as 32 P labeling of synthetic substrate (AKRRRLSSLRA) in the presence of PKA, PKC, and calcium-dependent kinase inhibitors using the S6 kinase assay kit (Upstate Biotechnology, Inc.) according to the manufacturer's instructions. Substrate-bound radioactivity was collected by binding to p81 phosphocellulose discs and quantified by scintillation counting after rinsing with phosphoric acid.
Endogenous or HA-tagged PKB/Akt was immunoprecipitated with rabbit polyclonal anti-Akt (Santa Cruz) or mouse monoclonal anti-HA (12CA5) (Boehringer Mannheim), respectively. For Western blotting, immunoprecipitated proteins were separated by 8% reduced SDS/ PAGE prior to transfer to nitrocellulose and analysis using the ECL kit (Amersham Pharmacia Biotech) with enhanced chemiluminescence. To measure kinase activity, immunoprecipitated PKB/Akt was reacted in kinase buffer containing [␥-P 32 ]ATP and histone 2B, as reported previously (36). 32 P-Labeled protein was separated by 15% SDS-PAGE and detected by autoradiography, and the relative intensity of labeling was quantified by phosphorimaging.

FIG. 1. Effects of PI3K inhibitors and rapamycin.
A, IGFBP-1 mRNA levels. HepG2 cells were stabilized in serum-free medium, then treated with 200 nM wortmannin (Wort), 50 M LY294002 (LY), or 200 nM rapamycin (Rapa) 30 min prior to the addition of 100 nM insulin or buffer alone (control). RNA was extracted 4 h after the addition of insulin and prepared for Northern blotting with an IGFBP-1 cDNA probe. Membranes were reprobed for GAPDH mRNA. Transcripts were visualized by autoradiography and hybridization intensity quantified by phosphorimaging. IGFBP-1 mRNA levels are adjusted for GAPDH and expressed relative to control. The mean and S.E. for a representative experiment performed in triplicate is shown. B, p70S6 kinase activity. HepG2 cells were stabilized in serum-free medium for 18 h, then treated with 200 nM rapamycin for 0, 4, or 20 h prior to the addition of 100 nM insulin. Lysates were prepared 10 min after the addition of insulin, and p70S6 kinase activity was measured after immunoprecipitation as the incorporation of 32 P into a synthetic peptide, as described under "Materials and Methods." C, IGFBP-1 promoter activity. HepG2 cells were transfected with 10 g of plasmid DNA, including 3 g of BP1.luc reporter gene plasmid, 2 g of ␤-gal vector, and 5 g of the ⌬p85 expression vector or empty vector. Cells were treated with 50 M LY294002 (LY), 200 nM rapamycin (Rapa), or Me 2 SO for 30 min prior to the addition of 100 nM insulin or buffer alone (control). Cell lysates were prepared 18 h later. Luciferase activity is adjusted for ␤-gal and expressed relative to control.

RESULTS AND DISCUSSION
As shown in Fig. 1A, insulin lowers IGFBP-1 mRNA levels in HepG2 cells by ϳ50%. Wortmannin, which directly inhibits the p110 catalytic subunit of PI3K, increases the level of IGFBP-1 mRNA and largely prevents the ability of insulin to lower IGFBP-1 mRNA levels. LY294002, another highly specific PI3K inhibitor that is structurally unrelated to wortmannin and more stable in solution, also increases IGFBP-1 mRNA levels and completely blocks the effect of insulin. In contrast, rapamycin reduces IGFBP-1 mRNA levels by ϳ20% and does not disrupt the ability of insulin to further lower mRNA levels.
In vitro kinase assays confirm that this dose of rapamycin reduces basal p70S6 kinase activity and blocks its activation by insulin for up to 20 h in HepG2 cells (Fig. 1B). These results indicate that PI3K contributes negatively to the regulation of basal IGFBP-1 mRNA levels and that insulin lowers IGFBP-1 mRNA levels through a PI3K-dependent mechanism that is not sensitive to inhibition by rapamycin.
We next asked whether PI3K is required for insulin to inhibit IGFBP-1 promoter activity. As shown in Fig. 1B, insulin inhibits IGFBP-1 promoter activity by ϳ50% in HepG2 cells, consistent with results obtained at the mRNA level. LY294002, which directly inhibits the activity of the p110 catalytic subunit of PI3K, increases IGFBP-1 promoter activity and prevents the inhibitory effect of insulin. Co-transfection with a dominant negative form of the 85-kDa regulatory subunit of PI3K (⌬p85) (27), which blocks the activation of PI3K by insulin without disrupting intrinsic p110 activity, also completely disrupts the inhibitory effect of insulin but without altering basal promoter activity. In contrast, treatment with rapamycin lowers basal promoter activity by ϳ20% and does not disrupt the inhibitory effect of insulin, consistent with results obtained at the mRNA level. These results indicate that PI3K contributes negatively to the regulation of basal IGFBP-1 promoter activity, and that insulin inhibits promoter activity through a PI3K-dependent mechanism that is not blocked by rapamycin.
To determine whether PI3K is sufficient to mediate the effect of insulin on promoter activity, we next performed studies with constitutively active forms of the 110-kDa catalytic subunit of PI3K created by joining the catalytic (p110) and regulatory (p85) subunits through a hinge peptide (p110*) (36) or by adding the Src myristoylation sequence to p110 (Myr-p110), as described previously (37). As shown in Fig. 2A, expression of increasing amounts of either p110* or Myr-p110 inhibits IG-FBP-1 promoter activity in a dose-dependent fashion. There is no additive effect of insulin on promoter activity in cells transfected with either p110* or Myr-p110, indicating that PI3K is sufficient to mediate the full effect of insulin on basal IGFBP-1 promoter activity.
We next asked whether PKB/Akt may mediate PI3K-dependent effects of insulin on the IGFBP-1 promoter. Western blotting and in vitro kinase studies demonstrate that PKB/Akt is expressed in HepG2 cells and activated by insulin (Fig. 3A). This effect of insulin is inhibited by LY294002, but not by rapamycin (Fig. 3A), consistent with previous studies regarding the activation of PKB/Akt (7).
To determine whether PKB/Akt is sufficient to mediate effects of insulin and/or PI3K on promoter activity, we next performed studies with constitutively active PKB/Akt (Myr-PKB/Akt). As shown in Fig. 3B, Myr-PKB/Akt inhibits IG-FBP-1 promoter activity in a dose-dependent fashion. This effect of Myr-PKB/Akt is not additive with the effect of either insulin or p110* (Fig. 3B), indicating that the effects of insulin and PI3K on IGFBP-1 promoter activity may be mediated through PKB/Akt. In separate experiments, 18-h conditioned medium collected from Myr-PKB/Akt-or p110*-transfected cells had no effect on IGFBP-1 promoter activity (data not shown), indicating that the effects of PKB/Akt and PI3K on promoter activity are due to stimulation of intracellular path- We next performed studies with the dominant negative K179 mutant of PKB/Akt (8). Initial dose-response studies showed that increasing amounts of the K179 PKB/Akt progressively block the effects of both insulin and p110* on IGFBP-1 promoter activity (not shown). As shown in Fig. 3C, K179 PKB/Akt disrupts the effects of insulin and PI3K on promoter activity without blocking the ability of Myr-PKB/Akt to signal downstream and inhibit the activity of the IGFBP-1 promoter. In vitro kinase assays and Western blotting reveal that K179 PKB/Akt prevents the activation of HA-tagged wild type PKB/ Akt by insulin and p110* without altering protein levels or disrupting the activity of Myr-PKB/Akt (Fig. 3C). These results indicate that insulin and PI3K act upstream of PKB/Akt and activation of PKB/Akt is required to mediate effects of insulin and PI3K on the IGFBP-1 promoter.
We next asked whether PI3K and PKB/Akt mediate their effects on promoter activity in a sequence-specific fashion through an IRS. As shown in Table I these results indicate that the effects of insulin, PI3K, and PKB/Akt are all mediated through IRSs in a sequence-specific fashion and that highly related IRSs from other hepatic genes also can mediate the effects of insulin, PI3K, and PKB/Akt on basal promoter activity in liver-derived cells.
These results provide the first report demonstrating that PKB/Akt exerts sequence-specific effects on gene expression and mediates the effect of insulin on basal hepatic gene expression through a conserved IRS. Early studies with cycloheximide revealed that insulin inhibits IGFBP-1 and PEPCK transcription through post-translational modification of preexisting factors (41,42). Meier et al. (43) recently reported that activated forms of protein kinase B are translocated to the nucleus where they may interact with and modify the activity of nuclear factors involved in the regulation of gene expression. Phosphorylation and inactivation of glycogen synthase kinase-3 (11), which is present in the nucleus and known to phosphorylate several transcription factors (44,45), may provide one mechanism by which PKB/Akt may alter gene expression. We previously reported that members of the HNF-3/forkactivation of PKB/Akt, cells were transfected with 12 g of DNA/dish, including 1 g of HA-tagged wild type PKB/Akt, 1 g of HA-tagged Myr-PKB/Akt, 1 g of p110*, 10 g of K179 PKB/Akt, and 10 -11 g of empty vector, then treated with 100 nM insulin or buffer alone for 10 min as indicated. Lysates were prepared in the presence of protease and phosphatase inhibitors and HA-tagged proteins precipitated for in vitro kinase assay (middle panel) or Western blotting (bottom panel), as described under "Materials and Methods."  (35,46). Subsequent studies in other laboratories have confirmed this result (47,48) and demonstrated that HNF-3/forkhead proteins also interact with the apoCIII IRS (48). An insulin-responsive sequence in the tyrosine aminotransferase gene also binds HNF-3/forkhead proteins (49). Additional studies are required to determine whether post-translational modification of HNF-3/forkhead proteins or other specific trans-acting factors mediate effects of insulin on hepatic gene expression downstream from PKB/Akt.
Based on these results, we suggest that, in addition to mediating effects of insulin on carbohydrate metabolism in peripheral tissues (5,6,50), PKB/Akt also plays a critical role in mediating effects of insulin on hepatic gene expression through a conserved IRS. Krook et al. (51) reported that the activation of PKB/Akt by insulin is impaired in skeletal muscle in insulinresistant diabetic rats. Bang et al. (52) reported that circulating levels of IGFBP-1 are not appropriately suppressed in insulinresistant patients with Type 2 diabetes and suggested that these individuals may be resistant to the effect of insulin on hepatic production of IGFBP-1. It remains to be determined whether defects in signaling through the PI3K-PKB/Akt-IRS pathway impair the ability of insulin to regulate hepatic function and contribute to the pathogenesis of some forms of diabetes mellitus.